Development of a Humanoid Robot Prototype for Multimodal Human-Robot Interaction

The description provided does not match any specific robot in the provided search context, which instead highlights the iCub (17-DoF upper body), Pepper (commercial social robot), NICOL (2-DoF head, 2x8-DoF arms), and the IQ-9100 (26-DoF).

• iCub: Features a 17-DoF upper body (2x7-DoF arms, 3-DoF torso) and is used in the HARMONIOUS framework for reactive motion control using vision, proximity, and tactile sensors.
• Pepper: A commercial humanoid used for multimodal emotion recognition integrating visual, auditory, and textual analysis.
• NICOL: Designed for collaboration, equipped with a 2-DoF head and two 8-DoF arms with five-fingered hands, grounded with LLMs and deep learning models.
• IQ-9100: An industrial/domestic humanoid with 26 controllable degrees of freedom and edge AI performance for real-time multimodal interaction.

The specific "14-DOF" prototype with an "expressive LCD head" mentioned in the query is not found in the provided search results.

The material presents the design and implementation of a 14-degree-of-freedom humanoid robot prototype intended as an open testbed for multimodal human-robot interaction. The platform combines dual-arm actuation with an expressive LCD-based head, enabling the robot to convey not only speech or text but also visual affective cues such as facial-like displays, gaze, and animated expressions. A custom controller is described as a central component for coordinating perception, decision-making, and actuation, allowing different AI modules to be integrated and evaluated within a shared hardware and software stack.

A key contribution is the framing of the robot as a modular research platform rather than a single fixed application. By providing a compact humanoid embodiment with expressive output channels and a controllable body, the system supports experiments spanning speech recognition, natural language understanding, affective computing, social signaling, and real-time motor control. This is especially relevant because many human-robot interaction studies are limited by proprietary hardware, fragmented software interfaces, or insufficient expressive bandwidth, making it difficult to compare algorithms across labs or to study how multiple interaction modalities influence user perception.

The work matters because it addresses a practical bottleneck in HRI research: the need for affordable, extensible, and reproducible embodied testbeds that can bridge high-level AI and physical interaction. A humanoid prototype with dual arms and a visual display head is well suited for studying tasks where body posture, arm motion, and facial feedback jointly shape trust, attention, and perceived social presence. As AI systems increasingly move from purely digital interfaces into embodied agents, open hardware platforms like this one can accelerate research into more natural, transparent, and human-centered robot behavior.